The URL https://platform.openai.com/docs/guides/prompt-engineering use to have a good overview of the guidelines, but this has replaced a much more unstructed text. Perhaps the old advice was too simplistic? Am I the only one to miss the old guidelines?
My view is that with the newer models such as GPT-5 and the more specific models for other tasks, one has to read the documentation for each, as what may work for one model is no longer required or a best practice for other models.